Ask anyone what AI means in 2025, and they’ll probably mention ChatGPT. Large language models (LLMs) have made this technology so accessible that the two have become synonymous. The same could be said for sophisticated chatbots like Claude, Copilot and Gemini. It’s become so easy to interact with these language models that a lot of people have started to use them as search engines. And why not? You get a direct response and don’t need to click through pages of sponsored web links to find the answer you’re looking for. But LLMs aren’t AI, they’re a kind of GenAI model. They’re fancy chatbots that create new content by predicting which words should come next in a sequence. And the content that comes out of an LLM is based on the vast amounts of data it’s been trained on.

The LLM ecosystem is massive. Hugging Face – an open-source platform for building and sharing AI models – currently has just under 1.5 million models. But most of the models aren’t LLMs because there is so much more to AI than text generation. There are computer vision models, audio processing models, multimodal LLMs, large data models, embedding models, diffusion models, classification models, among others.